Designing collaborative musical experiences for broad
audiences
Ben Bengler
Interactional Sound and Music
Centre for Digital Music
School of Electronic Engineering and Computer
Science
Queen Mary University of London
Mile End Road, London E1 4NS, UK
benedikt.bengler@eecs.qmul.ac.uk
ABSTRACT
This paper reports on the design and audience evaluation
of a collaborative interactive music system titled Polymet-
ros. Designed for broad audiences, Polymetros aims to en-
able users without musical skills to experience collaborative
music-making. First, we describe our design approach with
reference to related research. A particular interest was to in-
vestigate how to provide novices with individual musical con-
trol within a collaborative context. We then present an audi-
ence evaluation that was conducted during an exhibition at a
major art museum in the UK attended by large numbers of
the general public across the age range. The results lead us to
evaluate our design approach and reflect on implications for
facilitating collaborative music-making for broad audiences.
Furthermore, the findings provide interesting indications how
the context of a public exhibition setting affects the audience
interaction with such an interactive multi-player experience.
Author Keywords
Collaboration, interactive music systems, interaction design,
multi-user interfaces, public creativity, audience experience,
field study
ACM Classification Keywords
H.5.3. Group and Organization Interfaces: Evalua-
tion/methodology; H.5.5. Sound and Music Computing: Sys-
tems
General Terms
Human Factors; Design
INTRODUCTION
This paper discusses results from an ongoing research project
that investigates how to transfer the traditionally exclusive ex-
perience of musical collaboration to a much broader audience
Permission to make digital or hard copies of all or part of this work for
personal or classroom use is granted without fee provided that copies are not
made or distributed for profit or commercial advantage and that copies bear
this notice and the full citation on the first page. Copyrights for components
of this work owned by others than ACM must be honored. Abstracting with
credit is permitted. To copy otherwise, or republish, to post on servers or to
redistribute to lists, requires prior specific permission and/or a fee. Request
permissions from Permissions@acm.org.
C&C’13, June 17 - 20 2013, Sydney, NSW, Australia
Copyright 2013 ACM 978-1-4503-2150-1/13/06...$15.00.
234
Nick Bryan-Kinns
Interactional Sound and Music
Centre for Digital Music
School of Electronic Engineering and Computer
Science
Queen Mary University of London
Mile End Road, London E1 4NS, UK
nickbk@eecs.qmul.ac.uk
using interactive technology. In contrast to Western art mu-
sic, where collaboration mainly aims for a particular musical
result according to ‘high art’ aesthetic criteria, we are inter-
ested in the creative process of making music as an end in
itself. By considering music-making as an enjoyable group
activity rather than an expert profession the idea of collab-
orative musical experiences refers to enabling people with-
out particular musical training to experience being part of a
creative, collaborative musical process. Common attributes
characterising similar approaches in the existing literature in-
clude ‘open-ended’ [1] [5], ‘social’ [2] [20] and ‘democratic’
[17] [22]. As part of this project, we developed the interac-
tive music system Polymetros which is designed as an audi-
ence experience for exhibitions, festivals or public workshops
while serving as a flexible research tool to investigate and
evaluate our strategies under such real-word conditions.
Since such contexts are generally intended to reach out to a
wide public, our target audience is characterised by a broad
range of skills, experience, and age. However, particularly
with regard to musical abilities, it has to be considered that
large parts of the audience in a public context cannot be as-
sumed to have honed musical skills gained through formal
musical training such as learning a musical instrument. In the
following paper we refer to this group as musical novices.
Therefore, a particular interest was to investigate how to pro-
vide novices with opportunities to experience and maintain
control over their musical activities and how these contribute
to the collaboratively created music. This approach was based
on our research claim, informed by relevant literature, that
feeling part of a collaborative musical process is closely re-
lated to the sense of control each participant has over their
particular contribution. Moreover, we consider individual
control as an essential pre-condition for active musical col-
laboration and therefore as an essential design goal for inter-
active systems aiming at facilitating such experiences.
The paper will first outline how our general approach and the
particular design of the Polymetros system was informed by
related research and then present the results of an audience
evaluation conducted at a public exhibition event.
CONTEXT AND RELATED WORK
Reflecting on what constitutes the satisfaction of making mu-
sic, Atau Tanaka argues that the musicians’ satisfaction is di-
rectly related to the sense of how their actions manifest in the
resulting music. This includes the direct feedback of the in-
strument as well as “the identifiability a musician maintains
in feeling the contribution his part is making in an ensem-
ble” [17]. In accordance to the general idea of the ‘sense
of agency’ which refers to the sense of control over one’s
own actions [7], Tanaka calls these notations ‘sense of mu-
sical agency’ [17]. Considering this concept with regard to
a scenario involving musical novices clearly illustrates the
challenge we face when designing interactive technology that
aims at transferring such an experience to a broad audience:
Unlike an ensemble consisting of musicians who spent years
of training in order to control and shape their instrument’s
feedback into a distinctive voice within the ensemble con-
text, collaborative musical experiences can not rely on skills
and practices that need to be developed over a long period
of time. Therefore, interactive technology steps in to bridge
these technical demands by facilitating possibilities for musi-
cal control and interaction that are achievable for everyone.
A common strategy is the use of pre-produced musical mate-
rial: Instead of enabling the creation of musical content, the
system provides high-level control possibilities that allow the
participants to influence its musical output. Multi-person ap-
plications that have explored such an approach include Rob-
son’s Sound Toys [15], the Jam-O-Drum system [3] or MIT
Media Lab’s Musical Navigatrics [13]. Even though these
systems differ significantly in terms of appearance and used
sensor technology, they share the same basic principle: Each
participant’s interaction is mapped to a high-level parameter
such as volume or timbre allowing them to modify a partic-
ular element of a pre-composed sound set. A common cri-
tique of such an approach is that the user’s possibilities to
have an impact on the musical outcome are often strongly re-
stricted in favour of an intended, predetermined output. Sergi
Jordà argues that by “seeking to guarantee a complex or pre-
defined musical output, many of these [sound interactives]
do not give to their interactors more than a couple of bits to
play with” [10]. Also Weinberg criticises the deficiency of
many novice systems for not allowing their players to “truly
contribute meaningfully and creatively to the composition,
but rather they are only allowed to manipulate and control
pre-composed material” [19]. Moreover, it is questionable
whether such restricted possibilities of influence are sufficient
to convey a ‘sense of musical agency’ and the experience of
actually making music.
In accordance with his critique, Jordà directed the devel-
opment of the collaborative interactive tabletop system Re-
actable [11], which has became quite well-known over recent
years. The design goals were to create a “state-of-the-art in-
teractive music instrument” that allows its players total con-
trol without being based on pre-produced material or presets.
It should be suitable for skilled electronic musicians in con-
certs as well as for complete novices in installations without
instructions [11]. The Reactable relies on the metaphor of
a modular synthesiser whereas the different available mod-
235
ules are represented by tangible objects. These include audio
generators like oscillators or sample players, control modules
like LFOs (Low frequency oscillator) and sequencer objects
as well as audio effects such as filters or delays. The modules
can be dynamically patched together on the tabletop surface
while their particular properties, mutual dependencies and the
overall data flow are visualised via rear projection. In this
way, several people can collaboratively build up a modular
interactive musical environment for real time sound synthe-
sis. Being a highly versatile and powerful environment for
experts, we consider it as less appropriate for music making
among novices. Even if suitable to allow a novice to exper-
imentally explore electronic sound synthesis, a collaborative
creation undertaken by several players simultaneously seems
to require a profound understanding of the underlying prin-
ciples and various interdependencies among the available ob-
jects.
Unlike Reactable, Weinberg’s collaborative music systems
are explicitly designed for novices. Avoiding the use of pre-
recorded audio, the musical output of the Beatbug Network
[20] is entirely based on material created by the users dur-
ing their playing experience. Each player is provided with
their own bug-shaped, hand-held interface allowing them to
enter a short rhythmic pattern which is then sent to an other
player’s bug chosen at random. The receiving player can de-
cide whether to transform it by applying audio effects via
the bug’s bend-sensor antennae and pass the modified version
on or keep the received pattern playing as is and contribute
their own pattern. In a turn-based manner, all group members
are able to contribute their own phrase while maintaining the
transformations they like. Similar to Weinberg’s Squeezables
[21], the Beatbug Network is also based on the concept of in-
terconnected musical networks (IMNs) that allow players “to
influence, share, and shape each others’ music in real-time”
[20]. Due to the complexity of the interactions, both systems
were mainly suited for structured performances that follow
a pre-composed score being rehearsed in a series of guided
workshops rather than open-ended audience experiences in a
public context [19]. But Weinberg’s network-based approach
reveals an interesting perspective on the relevance of a per-
sonal contribution in such environments: Informed by his de-
sign experiences, Weinberg recommends restricting the level
of interdependency in order to avoid “uncertainty about the
individual control of each player” [21]. Therefore, a player
should not be able to modify a peer’s phrase beyond recog-
nition, as its creator might experience a loss of control and
influence on the overall musical result. This can make the
players “feel disconnected from the music they created, as
their detailed idiosyncratic contribution might be eliminated”
[19]. These claims closely correspond to Tanaka’s concept
of musical agency and highlights the importance of provid-
ing players with opportunities to create and maintain clearly
identifiable contributions within a collaborative musical con-
text.
POLYMETROS
We now come to a description of Polymetros, which is an in-
teractive, table-based music system designed as an audience
experience for 3 - 4 participants. Every player is provided
with their own physical user interface allowing for an indi-
vidual musical contribution within this collaborative context
(Figure 1).
Figure 1. Polymetros
Design strategy
In light of findings from our literature review as discussed
in the above section, we aimed to find a design approach
that is neither based on pre-produced musical material nor
on concepts or metaphors originating from a technical back-
ground. Therefore we considered an interaction concept re-
lying on basic musical properties such as note, pitch and
time. While the use of these commonly understood proper-
ties seems promising, it poses the challenge of how to en-
able novices to create their personal musical contribution on
a basic musical level while also allowing for collaboration
in such a way that manages the tension between ‘cacophonic
chaos’ and an oversimplicity caused by highly restricted pos-
sibilities. From assessing and refining this strategy, the idea
emerged to use Minimal Music as an inspirational basis for
the system design.
Referring to a musical style mainly associated with the Amer-
ican composers Steve Reich, Phillip Glass, Terry Riley and
La Monte Young [12], Minimal Music is characterised by the
use of short repetitive musical phrases that are gradual mod-
ified in length, time signature or melodic shape. This leads
to procedural transformations of the resulting music that of-
ten maintains a stable harmony while the main interest lies
in the perceptible process itself [14]. As described by Phillip
Glass, Minimal Music “no longer has a mediative function,
referring to something outside itself, but it rather embodies
itself without any mediation” [12]. This conception of music
without an imposed higher sense being rather an end in itself
closely relates to the idea of a musical experience as defined
earlier. The music qualities are often characterised as ‘hyp-
notic’, ‘trance-like’ causing a sense of timelessness. Beside
that, Minimal Music implies a ‘democratic’ attitude towards
players as well as musical material: There are no soloists or
lead instruments as no sound has any greater importance than
another [12].
As these qualities, the musical structure, and the underlying
techniques of Minimal Music closely corresponded to our in-
tended scenario, we decided to use specific properties of this
236
musical ‘expert-domain’ to inform the design of an interac-
tive system for a broad audience. In practical terms, we aimed
to allow for every player to create and control a repetitive mu-
sical phrase which would become their personal contribution
to a constantly evolving, collaborative musical output. Be-
sides editing the actual notes of the motif, they should be able
to dynamically change tempo and time signature, add and re-
move notes/rests or shifting their phrase by a metrical entity
in relation to their co-players’ contributions. Inspired by min-
imalist composing techniques, these interventions allow the
users to continuously deduce new variations and modify their
contributions in a way that results in distinct alterations in the
overall musical output.
However, it is important to note that this inspirational in-
fluence does not inevitably constrain the system’s output to
Minimal Music in the traditional sense. The derived control
strategies provide a wider, versatile range of possibilities in
terms of pattern-based music, for example in the studies re-
ported in this paper we observed participants using Polymet-
ros to create music almost in the style of Dubstep.
Figure 2. A Polymetros instrument
System description
Polymetros is based on a software application written in MAX
which controls a multi-user hardware interface combined
with external audio equipment for sound synthesis and au-
dio routing (Figure 3). The user interface is composed of 3-4
individual hardware devices which are connected to a cen-
tral hub in a star-like topology. Each participant controls one
device which features a grid of illuminated buttons. These
devices are referred to as instruments (Figure 2). In order to
implement the musical possibilities as described previously,
every player can create and edit their musical phrase via a 8
by 8 button matrix where the x direction corresponds to time
in beats and the y direction to pitch according to a chosen mu-
sical scale. In this case study a natural minor scale was used.
The phrase is continuously played back in a loop while the
actual playing position is indicated by a horizontal bar ‘trav-
elling’ over the grid. Via the button row on top of the melody
grid the user can dynamically change the loop length by defin-
ing the active section on the grid. The function buttons on the

Figure 3. Technical overview
right allow the user to shift the active section, change tempo
and octave register as well as mute the instrument’s output.
All functions and states of the interface are indicated by visual
feedback via tri-colour LEDs embedded in the buttons. In ad-
dition, the instrument’s cable connection to the hub flashes
up every time a note is triggered on its grid. The sphere-
shaped hub serves as the visual centre pulsating red with the
beat as the metaphorical ‘heart’ of the system. In the reported
study the sounds of the instruments were based on different
mallet and pitched percussion instruments as these sound rea-
sonably pleasing over a very wide pitch range. Polymetros is
assembled on a round table with built-in loudspeakers that
are hidden under the cloth stretched over the tabletop. These
are mounted in such a way that each speaker is directed at
one particular player serving as their personally assigned au-
dio monitor. Using high-quality miniature speakers with a
directional characteristic, an equal and well-defined listening
situation can be provided for all players.
CASE STUDY
Pilot
Before the case study presented here, a pilot was conducted
at an open studio event at the authors’ university as part of the
Digital Shoreditch Festival 2012. During this three-day exhi-
bition the initial version of Polymetros was played by around
300 people while being video recorded and attended by one
of the authors over the entire course of the event. The find-
ings significantly informed the practice and methodological
approach of how to conduct an ‘in-the-wild’ study of such a
system within the context of a public exhibition. Beside that,
they led to several improvements of the initial prototype in-
cluding system reliability and user interface design as well as
the implementation of often requested features such as indi-
vidual tempo control.
Context and setup
The user study was conducted during the exhibition of Poly-
metros at the Digital Design Weekend 2012 which took place
on September 22nd and 23rd at the Victoria and Albert Mu-
seum (V&A) in London, UK. Being part of the London De-
sign Festival, this annual event is dedicated to contemporary
237
digital art and design, including interactive installations, per-
formances, demonstrations and workshops. Incorporating an
open studio-like atmosphere, most of the exhibiting artists
attended the event in order to demonstrate and discuss their
work with audiences.
Polymetros was installed in a slightly darkened studio space
alongside several other interactive and static artworks that
required dimmed lighting. The system was positioned near
a corner of the room sufficiently far from the walls to be
approachable from all sides. Even though Polymetros was
the only exhibit that incorporated sound within this area, the
overall noise level was rather high due to the event’s open
studio character. On request of the curator the originally in-
tended output volume had to be lowered to ensure that the
sound is not audible in adjacent areas. Two video cameras
were installed to record the user interaction from two differ-
ent angles.
For conducting the case study we were aided by two assis-
tants. This allowed the onsite researcher to mainly focus
on contextual observation while the collaborators conducted
and coordinated the other study-related tasks. These involved
controlling the data logging system as described in the fol-
lowing section, handing out the questionnaires and providing
some initial guidance on the interface functionality. Due to
the open studio character of the event, the research team was
also available to answer specific questions and discuss the ex-
hibit with interested audience members.
Methods and Data collection
Informed by the pilot study, we applied a mixed-methods ap-
proach using questionnaires, interaction logs, field observa-
tions and video analysis taking into account the contextual
demands and the complexity of the setting.
Questionnaire
The questionnaire was designed to capture how participants
rate different aspects of their playing experience. It was
handed out by a member of the research team who ap-
proached visitors just after they finished playing with Poly-
metros with the request to fill in a questionnaire. Consider-
ing the given context, the questionnaire was designed to be
completed in a short amount of time (1-3min). Beside gath-
ering demographic data such as age and gender, it consisted
of nine statements the participants were asked to rate on a
5 point Likert-type scale (not at all - slightly - moderately
- fairly - very much). The questionnaire items are listed in
Table 1. While the first two statements referred to the par-
ticipants’ musical skills and their experience with interactive
technology (S1, S2), the other seven items related to the ac-
tual playing experience (E1-E7). During the two days of the
exhibition the questionnaire was filled in by 150 participants
(82 female and 68 male) from ages 5 to 66 (M = 24.3, SD =
14.05).
Data logging
For recording the user interaction with the interface a data
logging application has been integrated into the Polymetros
system. Via a remote control device the facilitator was able
to activate and stop the data recording according to the time
 S1: I am a musical person. STUDY RESULTS
S2: I am experienced using interactive devices.
(e.g. smart phone, tablet, video gaming)
S1: musicality
E1: I felt part of a creative process.
E2: I felt in control. S2: tech. experience

E3: I felt connected to the other players.

not at all slightly moderately fairly very much
E4: It was challenging.
E5: I liked the music we created. Figure 4. Self-rating of musicality and technical experience (150 partic-
ipants)
E6: My playing was influenced by the playing of the others.
E7: I would recommend playing Polymetros to my friends. The data of the questionnaire are shown in Figure 4 and Fig-
Table 1. Questionnaire statements. ure 5 . Each survey item is represented as an individual box
plot showing median (bold, horizontal line within the box),
a user spent playing on a particular instrument. Each created first and third quartile (edges of the box), extreme values
log contained an ID code indicating the number and instru- (whiskers) and outliners (empty dots) of the corresponding
ment used, the start and end time as well as all edits made dataset. Beside that, the mean value is marked with a dia-
with the interface. During the study the most recent two ID mond symbol.
codes of each instrument were shown on a display clearly vis-
ible for all members of the research team. This allowed the fa-
cilitator handing out the questionnaires to mark collected sur-
veys with the corresponding ID if a log file has been created E1: process
for this particular user. In this way, 63 of the 150 completed
questionnaires could be allocated to their corresponding set E2: control
of interaction data.
E3: connection
Field and video observations
The observations were carried out by one of the authors for E4: challange
several hours on both days of the exhibition. They were
E5: music
mainly conducted from a standing desk located in the corner
of the room which was also used to monitor the system func-
E6: influence
tionality via a nearby computer display. Descriptive notes
were taken referring to specific playing experiences of one or E7: satisfaction
several audience members as well as events and activities of
interest in the surrounding area. Beside participants’ actions,
0 all
not at 1
slightly 2
moderately 3
fairly very 4much
the records included related gestures and behaviours such as
laughing or head-nodding as well as age, gender and the cor-
responding ID code if available. Due to the close proximity Figure 5. Items related to playing experience (150 participants)
to the users, we were able to note down verbal statements
including spontaneous ‘outlouds’ of players, communication
between audience members and comments made towards the
research team. Experience questionnaire
While the self-rating of the participants’ musical abilities is
In order to complement the field notes and examine specific distributed over a wide range (S1), the majority of them con-
aspects in greater detail 5 hours of the recorded video material sidered themselves as experienced users of interactive tech-
were examined and annotated. In contrast to the field obser- nology (S2). Visual inspection of the experience-related
vations, studying the video recordings allowed us a more pre- items shows that most participants evaluated their playing ex-
cise observation of how participants used their instrument’s perience as very positive: Designed to rate their overall satis-
interface and what kind of musical contributions they cre- faction in an indirect way, 102 of 150 respondents stated they
ated. It also enabled us to observe and review activities of would very much recommend playing to their friends (E7). In
interest that unfolded over a longer period of time such as a addition, the majority of participants indicated that they were
person’s transition from an observer to an active player (cf. pleased with musical result (E5) and that they felt part of a
[16]) or details that were difficult to recognise on site due creative process (E1). Beside that, we note a relatively high
to the surrounding audience. Beside that, all relevant verbal rank of the item ‘feeling connected to the other players’ (E3,
data recorded by the camera-mounted microphone were tran- median = fairly).
scribed.
The high rating of the item referring to the players’ feeling
of control (E2) suggests that the chosen design approach is

238
generally appropriate for broad audiences. For exploring po-
tential associations between individual survey items we used
the Goodman and Kruskal’s gamma coefficient as a dedicated
measure of associations between ordinal variables with rela-
tively few categories. Due to our specific interest in how the
feeling of being in control is related to other qualities of the
experience we calculated gamma for several combinations.
We found highly significant associations between ‘feeling in
control’ and ‘feeling part of a creative process’ (gamma =
0.56, p < 0.0001), ‘feeling in control’ and ‘recommend play-
ing to friends’ (gamma = 0.50, p < 0.0001) and ‘feeling in
control’ and ‘feeling connected to other players’ (gamma =
0.43, p < 0.0001). These findings indicate that experienc-
ing control is related to other key qualities of the participants’
playing experience such as feeling part of a creative process
or feeling satisfied with the overall interaction.
With the term dwell time we refer to the period of time a par-
ticipant was actively engaged with one of the instruments.
Based on 294 interaction logs, the average dwell time was
3.3 minutes. We used the matched data set of questionnaire
data with associated interaction logs as described previously
to explore the relationship between the participants’ dwell
time and their survey responses to the items ‘process’ (E1),
‘control’ (E2), ‘like’ (E5), ‘satisfaction’ (E7). Calculating
the Spearman’s rank correlation coefficient for these com-
binations indicated no association between the dwell time
and how the respondents rated such aspects in the experience
questionnaire. This finding raised our interest in examining
what factors were likely to be relevant for the participants’
dwell time in our specific scenario and we devoted particular
attention to this issue during the video analysis.
Observations
Over the two day period of the exhibition, Polymetros demon-
strated a strong appeal to the audience: While most of the
time all three instruments were in use, the players were usu-
ally accompanied by a number of spectators surrounding the
system. This situation seemed to establish an area of inter-
est attracting further passing visitors to take a closer look at
Polymetros whereas the majority stayed to interact with it. A
similar social dynamic relating to interactives in public areas
was reported by Brignull and Rogers referred to as the ‘honey
pot effect’ [4].
In this rather crowded setting people appeared to start learn-
ing how to interact with the system by watching previous
players: Several audience members were observed perform-
ing well-directed actions immediately after they took over
one of the instruments, showing a prior understanding of
the interface. Only complementary features such as length
or tempo selection were left to be indicated by the facilita-
tor. In general, we found that most of the audience mem-
bers understood the interaction concept when given a short
guidance by the facilitator. However, due to the rush of peo-
ple over long periods of the exhibition it was not possible to
provide guidance to all newly arrived players. While many
visitors were able to discover the relevant aspects of their in-
strument on their own, a number of audience members could
be observed having difficulties in understanding the interface.
239
Figure 6. Typical contributions
These seemed to be caused by a lack of understanding of the
loop-based concept of the user interface, as suggested by ob-
servations of inappropriate input gestures in relation to the
sequential playback of the instruments grid.
A key experience for the participants appeared to be the mo-
ment of ‘getting-it’ which could be observed when a hint
given by the facilitator or a co-participant led to the under-
standing of how their instrument works. Accompanied by the
first realisation of what is actually played by their instrument,
many participants reacted to this understanding with a facial
expression of excitement or spontaneous outlouds. This could
be described as a particular kind of “Aha” moment (cf. [6]),
the “This is me!” experience.
A prevalent input strategy was the creation of musical pat-
terns characterised by simple geometric properties. The
most common phrases consisted of horizontal and upward or
downward diagonal lines whereas in most cases all available
notes were used (Figure 6). Resulting in ‘closed musical fig-
ures’, this approach was applied by many players providing a
clear audio-visual correlation between the representation on
the interface and the musical result. It appeared to us that this
preference was related to the fact that for many people such
figures were easier to identify in the overall musical outcome
compared to more ‘sparse’ or complex patterns. Often such
phrases were modified in an incremental manner by changing
just a single event per playback cycle. This strategy suggests
a systematic attempt to create distinct alterations that provide
clear feedback affirming of the user’s understanding of the
interface.
In several cases, the similar sound characteristics of the differ-
ent instruments appeared to cause difficulties for the players
in identifying their pattern in the overall musical output. The
reason for choosing either mallet or pitched percussion-based
sounds for all instruments was motivated by the fact that these
are well qualified to be transposed over a very wide pitch
range having a versatile sound characteristic: It ranges from
drum-like sounds in low registers to bright, bell-like sounds
in high registers. This design choice aimed to enable all play-
ers to modify their phrase without restrictions over a wide
pitch range. But as a consequence, the instruments sounded
quite similar when played in the same octave register. While
this finding suggests that it may be better to use more diverse
sounds where each is restricted to a certain pitch range, this
would come at some cost, as it appeared that many players
particularly enjoyed ‘moving’ their phrase over a wide oc-
tave range in a dynamic manner resulting in an immediate
and salient acoustic feedback.
Despite the fact that the interaction with the Polymetros ap-
peared to be attractive and very enjoyable for the exhibition
audience, it has to be noted that collaborations involving ac-
tive engagement between ‘instrumentalists’ could be rarely
observed during the case study. By active engagement we re-
fer to a situation where players coordinate their actions in a
systematic way in order to develop the musical outcome over
a certain period of time. However, we observed several at-
tempts between players who commented on their actions or
discussed their playing activities across the table. In all these
cases the participants appeared to know each other being ei-
ther friends, couples or family members.
Nevertheless, in most instances audience members who knew
each other collaborated by playing together on a single instru-
ment, rather than on multiple instruments. They were more
likely to explore Polymetros together by explaining the inter-
face to their companion, co-editing a musical pattern or com-
menting on each others’ actions. These observations correlate
to the findings of von Lehn, Heath and Hindmarsh [18] show-
ing that interactive exhibits are often examined by visitors in
interaction with their companions. Such co-participation on
a shared instrument appeared to be a highly collaborative ac-
tivity in itself.
But unlike in the pilot study, where several groups were ac-
tively ‘performing’ together, such attempts involving all play-
ers could not be observed during the case study. Reviewing
the video material of the pilot showed that such actively co-
ordinated collaborations mainly took place if audience mem-
bers who were already acquainted approached the Polymet-
ros system whilst it was not in use. In such cases it appears
most likely that a group splits up to play on different instru-
ments, explores the system together and develops strategies
to coordinate their actions via verbal and non-verbal commu-
nication. In addition, this approach seems to promote mutual
awareness and interest for the co-players’ actions. But due
to the high visitor turnout during the case study at the V&A,
new players usually joined in an ongoing musical process in-
dividually rather than exploring the system together. Most of
all, not knowing the other players appeared to be the main
barrier to actively engage with them in order to jointly co-
ordinate the musical outcome. Although it could be noticed
that players were observing other participants’ actions or dis-
cussing their own playing with their companions or members
of the research team, they would not directly address their
unknown co-players. However, playing techniques such as
muting one’s instrument rhythmically in relation to other pat-
terns, which could be observed several times during the study,
showed an explicit awareness for other players’ contributions.
240
Beside that, the high visitor turnout during the study seemed
to have a significant influence on how long the visitors stayed
to play with the system: The video observations indicate that
people are likely to leave their instrument if they become
aware that another audience member is waiting in their di-
rect vicinity. It appears that in such a situation many audi-
ence members feel an ‘obligation to leave’ in order to make
room for other visitors. This is supported by the fact, that the
average dwell time of 3.3 min (294 logs) approximately cor-
responds to the dwell time measured on the opening evening
of the pilot study, where the system was similarly highly fre-
quented (3.8 min, 92 logs). On the two following days of
the pilot, which were much less well attended, the average
dwell time increased to 6 min (72 logs). Along with the ob-
servation that several visitors seemed to quit playing merely
because their companions were moving on, this suggests that
in many cases the dwell time was determined by contextual
factors rather than the participants’ playing experience. This
is supported by the results presented earlier indicating no as-
sociation between the participants’ dwell time and how they
rated their playing experience in the questionnaire.
REFLEXIONS
In the following section we reflect on the results of the study
in order to evaluate our design approach and assess implica-
tions for facilitating collaborative musical experiences.
On a general level, the study indicated that the players’ ‘sense
of control’ is an important feature of a collaborative musi-
cal experience and relates to other important qualities such as
the participants’ overall satisfaction with their playing expe-
rience. Our approach to facilitate such a feeling of control
through enabling each player to contribute on a basic musical
level drawn from Minimal Music appeared to be an appro-
priate strategy for an interactive collaborative music system.
Avoiding the use of pre-produced audio, Polymetros demon-
strated that it is understandable and engaging for a diverse
user group with different musical and technical abilities al-
lowing all players to contribute individually to an overall mu-
sical outcome. However, the observed difficulties of several
players in identifying the own contribution which was caused
by the similar sounding instruments, suggest opportunities for
a re-design of the sonic properties. One possibility is to im-
plement more idiosyncratic sound characteristics for each in-
strument. On the other hand, such an approach has to be care-
fully assessed taking into consideration that it might counter-
act the players’ individual control over their contributions by
restricting the variability of their instruments’ sound charac-
teristics. However, an important feature of the system is that
the sound palette can easily be changed for different contexts,
for example using drum, bass and synthesiser sounds ‘shifts’
Polymetros’s musical character into the realm of electronica
and club music.
In contrast to screen-based interactive exhibits, which have
been criticised for excluding the surrounding audience from
understanding the interaction with the system by undermin-
ing “mutual and public visibility of conduct” [8], the physical
interface approach showed to be an appropriate design choice
for the multi-user scenario: Providing a highly visible repre-
sentation of the interaction, the design demonstrated that it
succeeded in facilitating understandability for the surround-
ing audience while giving the participants visual access to
their co-players’ actions. Beside that, the physical and tactile
properties of the instruments enabled more than one person to
use a single instance of the user interface simultaneously - an
opportunity that is often entirely disregarded by screen-based
interactives [9].
As indicated by the observations, especially contextual influ-
ences such as unfamiliarity with other players and the fact that
visitors mostly joined in individually rather than exploring
the system as a group appeared to restrict collaborations in-
volving active engagement between audience members. Re-
flecting on these findings made us aware that from a visitor’s
perspective ‘active collaboration’ might not be a necessarily
relevant or desirable aspect in the context of such a highly-
frequented, public setting. This corresponds to findings of
von Lehn et al. indicating that it rarely occurs that exhibits in
galleries or museums are jointly explored by strangers, even
if these are designed to facilitate co-participation or collab-
oration [18]. However, it is interesting to note that, despite
the limited degree of active engagements between different
‘instrumentalists’ during the study, half of the respondents
stated in the questionnaire that they felt either fairly (34.7%)
or very much (16%) connected to the other players. This indi-
cates that many audience members experienced their playing
as a joint activity.
Nevertheless, based on our experiences from the pilot study
and informal evaluations we believe that active engagement
among players has the potential to significantly deepen the
playing experience for all participants. This raised our inter-
est in how we can more successfully facilitate collaborations
involving more active engagement while ‘reducing’ the ob-
served contextual influences. We explored an intuitive strat-
egy of facilitation within the last hours of the exhibition, when
Polymetros was less attended. Instead of solely providing an
initial guidance to each player individually, the facilitator in-
vited all present players to explore the system together: First,
group members were asked to ‘wipe’ the grids of their in-
struments in order to start a ‘new composition’ together. By
giving instructions to the whole group the facilitator encour-
aged the players to contribute one after another in a round-
based manner in order to successively build up the musical
outcome together. This approach appeared to support mutual
awareness among players by guiding their focus within the
group. Moreover, it clearly illustrated ‘who is playing what’
and how the overall musical outcome emerges from the indi-
vidual contributions of each participant. Thus, we suggest, it
appeared to promote a deeper overall understanding for func-
tionality and collaborative possibilities of the system for both
players and spectators.
FUTURE WORK
From our experiences described in the previous section, we
are interested in exploring this initial group facilitation strat-
egy in future exhibitions of the system. Furthermore, we plan
to complement this approach in a more controlled study set-
ting such as lab-based group activities in which players are
241
encouraged to compose and perform little musical pieces to-
gether whilst manipulating features of the system to study the
influence of individual control. However, as the main interest
of our research lies in the support of public musical creativity,
we are aware that results from a controlled study have to be
carefully balanced with the findings of our field research due
to the significant contextual influences as observed in the pre-
sented study. Besides, we are currently further analysing the
gathered interaction data in order to characterise the musical
contributions created by the participants and investigate how
these relate to their responses given in the questionnaire using
the matched data set.
SUMMARY
In this paper we reported on the design and audience eval-
uation of a collaborative interactive music system for broad
audiences. We first gave a detailed description of the design
rationale in relation to relevant literature with a particular fo-
cus on how to provide novices with opportunities for musical
control in collaborative contexts. The resulting design en-
ables every player to create and control a repetitive musical
pattern inspired by minimalist composing techniques aiming
to support a clear identification of their playing in the col-
laboratively created music. The case study emphasised the
importance of supporting each player’s sense of control and
indicated that it relates to other qualities of the participants’
playing experience such as feeling satisfied with the overall
interaction or feeling part of a creative process. In terms of
design, the findings suggested improvements of the instru-
ments’ sound characteristics to enhance their identifiability
in the overall musical outcome. We also explored an initial
strategy of facilitation in order to encourage active engage-
ment among the players.
The presented field study gave clear indications that the ways
the audience approached, explored and interacted with Poly-
metros were significantly influenced by the context of the
public exhibition. While the observations are specific to the
case study reported in this paper, we believe that, at a more
general level, they contribute to the understanding of audi-
ence interaction with collaborative interactive experiences in
public settings and are relevant to inform other designs that
aim to facilitate collaborative public creativity.
ACKNOWLEDGMENTS
We would like to thank Nela Brown, Kareem Khazem, Sara
Heitlinger, Richard Kelly, Colin Powell, Mark Plumbley and
Irini Papadimitriou.
This work was kindly supported by the Centre for Digital Mu-
sic (C4DM), Queen Mary University of London and the EP-
SRC funded DePIC project (EP/J017205/1), Benedikt Ben-
gler is funded by a PhD Studentship from Queen Mary Uni-
versity of London.
REFERENCES
1. Blaine, T. New music for the masses. Adobe Design
Center, Think Tank Online http://www.adobe.com/uk/
designcenter/thinktank/ttap_music/ (accessed 21
September 2012) (2006).
 2. Blaine, T., and Fels, S. Contexts of collaborative musical 16. Sheridan, J., and Bryan-Kinns, N. Designing for
experiences. In Proceedings of the 2003 conference on performative tangible interaction. International Journal
New interfaces for musical expression, NIME ’03, of Arts and Technology 1, 3-4 (2008), 288–308.
National University of Singapore (Singapore, Singapore,
17. Tanaka, A. Interaction, experience and the future of
2003), 129–134.
music. In Consuming Music Together, B. O’Hara,
3. Blaine, T., and Perkis, T. The jam-o-drum interactive K.; Brown, Ed., vol. 35 of Computer Supported
music system: A study in interaction design. In Cooperative Work. Springer Netherlands, 2006,
Proceedings of the 3rd conference on Designing 267–288.
interactive systems: processes, practices, methods, and
18. von Lehn, D., Heath, C., and J., H. Exhibiting
techniques, DIS ’00, ACM (New York, NY, USA, 2000),
interaction: Conduct and collaboration in museums and
165–173.
galleries. Symbolic Interaction 24 (2001), 189–216.
4. Brignull, H., and Rogers, Y. Enticing people to interact
19. Weinberg, G. Interconnected Musical Networks -
with large public displays in public spaces. In In
Bringing Expression and Thoughtfulness to
Proceedings of the IFIP International Conference on
Collaborative Group Playing. PhD thesis, MIT, 2003.
Human-Computer Interaction (INTERACT 2003)
(2003), 17–24. 20. Weinberg, G., Aimi, R., and Jennings, K. The beatbug
network: A rhythmic system for interdependent group
5. Bryan-Kinns, N., and Hamilton, F. Identifying mutual
collaboration. In Proceedings of NIME ’02 (2002),
engagement. Behaviour Information Technology (2009),
107–111.
1–25.
21. Weinberg, G.; Gan, S. The squeezables: Toward an
6. Csı́kszentmihàlyi, M. Creativity: Flow and the
expressive and interdependent multi-player musical
Psychology of Discovery and Invention. Harper
instrument. Computer Music Journal 25, 2 (July 2001),
Perennial, 1996.
37–45.
7. Gallagher, S. Philosophical conceptions of the self:
22. Xambó, A., Laney, R., and Dobbyn, C. Touchtr4ck:
implications for cognitive science. Trends in Cognitive
Democratic collaborative music. In Proceedings of the
Sciences 4 (2000), 14–21.
fifth international conference on Tangible, embedded,
8. Heath, C., Luff, P., von Lehn, D., Hindmarsh, J., and and embodied interaction, TEI ’11, ACM (New York,
Cleverly, J. Crafting participation: designing ecologies, NY, USA, 2011), 309–312.
configuring experience. Visual Communication 1 (2002),
9–33.
9. Heath, C., and von Lehn, D. Configuring ‘interactivity’:
Enhancing engagement in science centres and museums.
Social Studies of Science 38 (2008), 63–91.
10. Jordà, S. Digital instruments and players: part i —
efficiency and apprenticeship. In Proceedings of the
2004 conference on New interfaces for musical
expression, NIME ’04, National University of Singapore
(Singapore, Singapore, 2004), 59–63.
11. Jordà, S., Kaltenbrunner, M., Geiger, G., and Bencina,
R. The reactable. In Proceedings of the International
Computer Music Conference (ICMC 2005) (2005),
579–582.
12. Mertens, M. American Minimal Music. Kahn & Averill,
1983.
13. Pardue, L. S.; Paradiso, J. A. Musical navigatrics: new
musical interactions with passive magnetic tags. In
Proceedings of the 2002 conference on New interfaces
for musical expression, NIME ’02, National University
of Singapore (Singapore, Singapore, 2002), 1–3.
14. Reich, S. Music as a gradual procces, reprinted in:
Writings on Music 1965-2000. Oxford University Press
2000, 1968.
15. Robson, D. Play!: Sound toys for non-musicians.
Computer Music Journal 26, 3 (2002), 50–61.

242